Reactive power throwing and step load collaborative test system for network-related test of phase modifier unit

Abstract

The invention relates to the technical field of high-voltage power transmission and phase modifier testing, in particular to a reactive power throwing and step load collaborative testing system for a phase modifier set grid-related test, and the system comprises a power grid simulation device which is used for simulating the basic current supply of an AC power grid and is connected to a PCC bus through a power transmission line impedance; the fault injection device is connected to the PCC bus and used for simulating a three-phase short circuit generated by equipment connected to the PCC bus so as to generate a short circuit signal; the power step device is used for adjusting the step amplitude to generate an amplitude signal; according to the technical scheme provided by the invention, the fault injection device and the power step device are guided by the control module to randomly inject a cooperative abrupt change signal (such as a combination of a three-phase short circuit and a load step) to the PCC bus; and possible and unpredictable concurrent or continuous disturbance working conditions of the phase modifier unit in actual complex power grid operation can be simulated in a high-fidelity manner.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of high-voltage power transmission and phase modifier testing, in particular to a coordinated test system for reactive power shedding and step load of grid-connected phase modifier unit test. BACKGROUND

[0002] As an important dynamic reactive power support and voltage stability device of modern power grid, the grid-connected operation performance of the phase modifier unit is directly related to the safety and stability of the power system. The grid-connected test is a key link to verify whether the phase modifier unit can meet the design function, dynamic response characteristics and safe and stable operation requirements under the real power grid environment. Through strict grid-connected test, it can ensure that the unit can effectively provide or absorb reactive power, quickly adjust voltage, and ensure the reliability and power quality of the power grid when the power grid is disturbed.

[0003] At present, the grid-connected test of the phase modifier unit generally adopts the method of independent test and analysis of a single typical working condition. For example, the "reactive power shedding" test (simulating the sudden loss of part or all reactive load of the unit) and the "step load change" test (simulating the rapid step change of the grid load or instruction) are carried out respectively, and the key parameter responses of the unit under such single disturbance are mainly focused on, such as the fluctuation amplitude, recovery time and overshoot of the terminal voltage, reactive power, speed, current and other indicators, so as to evaluate the static and dynamic response efficiency of the unit.

[0004] The above-mentioned separate test method has significant limitations. It cannot effectively simulate and track the actual working process of the phase modifier unit in the complex power grid environment, which may face multiple concurrent or rapid switching disturbances. This fragmented testing method cannot fully and truly reflect the cooperative working ability and potential interaction problems of the unit control system, excitation system, protection system and other systems under dynamic coupling and continuous change conditions, and may cover up the deep-seated problems such as control logic defects, improper parameter setting, hidden faults of components or poor coordination between systems in the unit, resulting in that the evaluation result of the actual grid-connected operation performance of the unit is inconsistent with the actual situation. SUMMARY

[0005] The summary part of the present application is used to introduce the concept in a brief form, which will be described in detail in the specific embodiment part. The summary part of the present application is not intended to identify the key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0006] Some embodiments of the present application propose a coordinated test system for reactive power shedding and step load of grid-connected phase modifier unit test to solve the technical problems mentioned in the background part.

[0007] As a first aspect of the present application, some embodiments of the present application provide a reactive power rejection and step load coordinated test system for phase modulation machine group grid involvement test, comprising: a power grid simulation device for simulating a basic current supply of an alternating current power grid and connected to a PCC bus through a transmission line impedance; a fault injection device connected to the PCC bus for simulating a three-phase short circuit generated by a device connected to the PCC bus to generate a short circuit signal; a power step device for adjusting a step amplitude to generate an amplitude signal; a control module for controlling the fault injection device and the power step device to randomly inject a sudden change signal to the PCC bus; a to-be-tested phase modulation machine group connected to the PCC bus for maintaining stable power transmission on the PCC bus in response to the sudden change signal injected by the fault injection device and the power step device; an information extraction device for extracting static performance indicators, dynamic performance indicators generated by the to-be-tested phase modulation machine group on the PCC bus in response to the sudden change signal, and working performance indicators of the to-be-tested phase modulation machine group; a monitoring device for inputting the static performance indicators, the dynamic performance indicators, and the working performance indicators to a test large model to generate a performance report of the to-be-tested phase modulation machine group.

[0008] The technical solution provided by the present application can simulate the unpredictable concurrent or continuous disturbance conditions that the phase modulation machine group may encounter in actual complex power grid operation by guiding the fault injection device and the power step device to randomly inject a coordinated sudden change signal (such as a combination of three-phase short circuit and load step) to the PCC bus. At the same time, the system synchronously collects and fuses the static performance indicators, dynamic performance indicators, and key operating parameters generated by the to-be-tested phase modulation machine group when responding to these coordinated disturbances, thereby realizing multi-dimensional and coordinated performance evaluation. This analysis method based on real complex condition simulation and multi-source data fusion can comprehensively and deeply reveal the actual dynamic response characteristics, control strategy effectiveness, and potential weak links of the phase modulation machine group, thereby significantly improving the accuracy and reliability of the comprehensive performance evaluation of the phase modulation machine group in actual application scenarios, and generating a performance diagnosis report that accurately reflects the real operating capacity of the phase modulation machine group and has high engineering guidance value.

[0009] Further, the static performance indicators include reactive power regulation range, reactive power regulation accuracy, short circuit ratio improvement amount, and filter distortion rate; The dynamic performance indicators include voltage recovery time, reactive power response time, short circuit current peak value, and DC power recovery time; The working performance indicators are abnormal features analyzed from original operating parameters, including rotor speed, voltage, current, torque, and magnetic flux.

[0010] The application cooperates with the injection of the reactive power and the step load composite disturbance signal to the PCC bus through the control module, simulates the complex working condition of the power grid with high fidelity; synchronously collects and fuses three types of key data representing the steady-state regulation capability boundary (static indicators: reactive power regulation range / accuracy, short-circuit ratio improvement amount, filter distortion rate), transient response quality (dynamic indicators: voltage / reactive power response time, short-circuit current peak value, DC power recovery time) and the operating state of the unit body (operating indicators: rotor speed, voltage, current, torque, magnetic flux). The cooperative test method can start from the state of the unit itself, combine its steady-state regulation capability and transient response characteristics, and comprehensively and finely evaluate the comprehensive performance, control strategy effectiveness and potential weak links of the phase-modulating unit, and reveal its limit capability and operation risk under real complex disturbance.

[0011] Further, the static performance indicators obtain the static information group under the single burst signal mode; The dynamic performance indicators obtain the dynamic information group under the multi-burst signal mode; The single burst signal is only an amplitude signal or a short-circuit signal, and the multi-burst signal is a composite signal of the amplitude signal and the short-circuit signal.

[0012] The application tests accurately in different modes: under the single burst signal mode (only amplitude step or only three-phase short-circuit), the static performance indicators (reactive power regulation range / accuracy, short-circuit ratio improvement amount, filter distortion rate) are obtained, the dynamic interference is effectively stripped, and the steady-state regulation capability boundary and inherent static characteristics of the phase-modulating unit are accurately measured; under the multi-burst signal mode (amplitude step and three-phase short-circuit composite disturbance), the dynamic performance indicators (voltage / reactive power response time, short-circuit current peak value, DC power recovery time) are obtained, the complex concurrent disturbance scene of the power grid is truly reproduced, and the dynamic response quality, support strength and stability of the unit in the severe transient process are comprehensively examined.

[0013] Further, a plurality of groups of multi-burst signals are preconfigured, each group of multi-burst signals is loaded in sequence, and the dynamic indicators and the working performance indicators are synchronously measured; The collected dynamic performance indicators are taken as the dynamic information group, and the collected working performance indicators are taken as the working performance group.

[0014] The application pre-configures multiple groups of composite disturbance sequences and loads them in sequence. When each group of composite disturbances (multi-burst signals) is applied, dynamic performance indicators (voltage / reactive power response time, short-circuit current peak, DC power recovery time) and operating performance indicators (rotor speed, voltage, current, torque, magnetic flux) are synchronously collected to form dynamic data sets and operating state data sets, respectively. In this way, the application deeply couples the external dynamic response characteristics of the unit and the internal key operating state, significantly strengthening the temporal and spatial correlation between the two types of data. Through the correlation analysis of these synchronous data, the internal relationship and causal relationship between the dynamic performance (such as response delay, overshoot, oscillation) and the potential defects or abnormal states (such as rotor overheating, magnetic circuit saturation, torque fluctuation, and insulation deterioration) of the phase-modulating unit can be effectively revealed, providing a key basis for accurately locating performance bottlenecks, diagnosing hidden faults, and optimizing control strategies.

[0015] Further, the pre-configured multiple groups of multi-burst signals are fault characteristic signals measured in the actual working environment of the phase-modulating unit to be tested.

[0016] The application directly uses the fault characteristic signals measured in the target power grid environment as the pre-configured multi-burst signal sequence, which maximizes the restoration of the representative complex disturbance patterns and their timing characteristics that the phase-modulating unit to be tested may encounter in future actual service scenarios. Under the excitation of this highly realistic disturbance sequence, the synchronously collected dynamic performance indicators and operating indicators can accurately reflect the real disturbance resistance, dynamic response characteristics, and potential operating risks of the unit in a specific application environment, significantly improving the accuracy of performance evaluation.

[0017] Further, the test large model includes: a performance processing sub-model that generates abnormal state probability information of the phase-modulating unit according to input abnormal characteristics; a dynamic performance processing sub-model that generates first hidden information based on the double-channel input abnormal state probability information and dynamic performance indicators; a static performance processing sub-model that generates second hidden information based on static performance indicators; a performance report generation model that generates a performance report of the phase-modulating unit to be tested according to the first hidden information and the second hidden information; the performance report includes the probability of the operating stability of the phase-modulating unit to be tested.

[0018] The application performs multi-level sub-model collaborative analysis: a performance processing sub-model quantifies an abnormal state probability, a dynamic performance processing sub-model fuses the probability information and dynamic indexes (voltage / reactive power response time, etc.), and generates implicit dynamic risk characteristics (first implicit information); a static performance processing sub-model analyzes static indexes (reactive power regulation range, etc.), and generates implicit steady-state capability characteristics (second implicit information). A performance report generation model comprehensively integrates the above implicit information, deeply excavates the coupling relationship and potential fault chain between dynamic and static states, and finally generates a performance report containing a quantified operation stability probability.

[0019] There are a large number of noise signals in the original operation parameters, especially for the phase modulation machine group, which is a normal device itself, and internal defects generally belong to hidden defects difficult to detect, assembly errors allowed by the process, etc. These are difficult to detect and difficult to represent when the phase modulation machine group is running, and most of them are mixed in the normal operation signals of the phase modulation machine group. Therefore, for fault detection, the signal-to-noise ratio of the extracted original operation parameters is extremely low, and inputting into the neural network model for training can only pollute the data and cannot guide the model training. Based on this, the application provides the following technical solutions: Further, the optimal penalty factor and modal decomposition number are selected from a global perspective based on a particle swarm algorithm, and the original operation parameters are decomposed and the abnormal characteristics are reconstructed by using the optimal component factor and modal decomposition number.

[0020] In the technical solution provided by the application, the VMD modal decomposition and reconstruction method is optimized by using a particle swarm algorithm, which can decompose the extracted original operation parameters into a large number of IMF components, and by screening out appropriate IMF components for signal reconstruction, a large number of irrelevant signal characteristics can be eliminated, thereby enhancing the characteristics.

[0021] Further, the abnormal characteristic extraction method includes the following steps: S1: obtaining all original operation parameters , t = 1, 2, … N; i = 1, 2 … 5, t is the index of the sampling point, N is the total number of sampling points, and i represents the index of the original operation parameter.

[0022] , , … are respectively rotor speed, voltage, current, torque and magnetic flux.

[0023] S2: performing VMD signal decomposition on each original operation parameter respectively; S3: correcting the penalty factor and modal decomposition number of each when performing VMD signal decomposition by using a particle swarm algorithm, so that each The penalty factor and the number of modal decompositions are obtained when the VMD signal decomposition is performed to minimize the envelope entropy. S4: parameters are adopted The penalty factor and the number of modal decompositions are obtained when the VMD signal decomposition is performed to minimize the envelope entropy. The VMD signal decomposition is performed respectively to obtain the respective IMF signal components; S5: feature screening is performed on the IMF signal components, and then the screened IMF components are reconstructed to obtain the respective The corresponding reconstructed signals .

[0024] In the technical scheme provided in the present application, when the original operating parameters are reconstructed, in order to obtain the optimal penalty factor and the number of modal decompositions, the minimum envelope entropy is taken as the loss function of the particle swarm algorithm. The minimum envelope entropy can measure the significance of the signal impact characteristics, that is, some signals with impact characteristics can be accurately extracted from complex noise signals. These signals are generally some specific signals generated by the internal rotor and stator of the phase modifier set during operation. In this way, the diffusion direction of the particle swarm is guided by the minimum envelope entropy, and the optimal signal structure mode can be obtained, thereby improving the signal-to-noise ratio after signal reconstruction. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 It is a structural schematic diagram of a coordinated test system for the reactive power rejection and step load of the phase modifier set involved in the test.

[0026] Figure 2 It is a structural schematic diagram of a test large model.

[0027] Figure 3 It is a structural schematic diagram of a sub-model in the test large model.

[0028] Figure 4 It is a logic diagram of the particle swarm algorithm update. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical scheme and advantages of the present application clearer, the technical scheme of the present application will be described clearly and completely in combination with the specific embodiments. The same reference signs in the drawings represent the same parts. It should be noted that the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the described embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0030] Compared to the embodiments shown in the drawings, feasible implementations within the scope of the present application can have fewer components, other components not shown in the drawings, different components, differently arranged components, or differently connected components, etc. In addition, two or more components in the drawings can be implemented in a single component, or a single component shown in the drawings can be implemented as multiple separate components.

[0031] Unless otherwise defined, technical terms or scientific terms used herein shall have the ordinary meanings as understood by one of ordinary skill in the art to which this application pertains. The terms "first", "second", and similar terms as used in the specification and claims herein do not necessarily have any order or sequence or importance, but are used to distinguish different components. Similarly, the terms "one", "another", and similar terms do not necessarily have any quantity limitation. "Upper", "lower", and similar terms are used only to indicate relative positions, and when the absolute positions of the described objects are changed, the relative positions can also be changed accordingly.

[0032] Reference Figure 1 , Embodiment 1: The reactive power shedding and step load cooperative test system for the grid test of the phase modifier unit includes a power grid simulation device, a fault injection device, a power step device, a control module, a to-be-tested phase modifier unit, an information extraction device, and a monitoring device. The power grid simulation device, the fault injection device, and the to-be-tested phase modifier unit are connected to the PCC bus. The control module is connected to the fault injection device and the power step device. The information extraction device is connected to the PCC bus, the to-be-tested phase modifier unit, and the control module, and the monitoring device is connected to the information extraction device.

[0033] The power grid simulation device is configured to simulate the basic current supply of the AC power grid and is connected to the PCC bus through the transmission line impedance. The fault injection device is connected to the PCC bus and is configured to simulate the three-phase short circuit generated by the device connected to the PCC bus to generate a short circuit signal. The power step device is configured to adjust the step amplitude to generate an amplitude signal. The control module is configured to control the fault injection device and the power step device to randomly inject a sudden change signal to the PCC bus. The to-be-tested phase modifier unit is connected to the PCC bus and is configured to respond to the sudden change signal injected to the PCC bus by the fault injection device and the power step device to maintain stable power transmission of the PCC bus. The information extraction device is configured to extract the static performance indicators and the dynamic performance indicators generated by the to-be-tested phase modifier unit on the PCC bus in response to the sudden change signal, as well as the working performance indicators of the to-be-tested phase modifier unit. The monitoring device is configured to input the static performance indicators, the dynamic performance indicators, and the working performance indicators to a test large model to generate a performance report of the to-be-tested phase modifier unit.

[0034] The power grid simulation device, the fault injection device, and the power step device constitute a scheme of the test environment, which is prior art and will not be described in detail here.

[0035] The key of the embodiment is that static performance indicators, dynamic performance indicators and working performance indicators are collected, the three indicators with correlation are trained in a test large model, and thus a performance report of the phase-modulation machine set to be tested is obtained. How to collect the static performance indicators, dynamic performance indicators and working performance indicators is correspondingly provided below.

[0036] The information extraction device comprises a static information collection module, a dynamic information collection module and a working information collection module. The static information collection module, the dynamic information collection module and the working information collection module are respectively used for collecting static performance indicators, dynamic performance indicators and working performance indicators.

[0037] The static information collection module is used for collecting static parameters on the PCC bus to obtain static performance indicators; the static performance indicators include a reactive power regulation range, a reactive power regulation accuracy, a short-circuit ratio improvement amount and a filter distortion rate. The reactive power regulation range is (Qmax-Qmin) / Qmax, , Qmin is a minimum value of the reactive power regulation range, Qmax is a maximum value of the reactive power regulation range; the reactive power regulation range is actually measured by continuously approaching the upper limit and the lower limit of the reactive power regulation of the phase-modulation machine set.

[0038] The reactive power regulation accuracy is ; ; ; ; wherein, Qref is a reactive power instruction value, Qmean is an average value of an actual reactive power, Qmax is a maximum continuous reactive power output capability, V represents a PCC bus voltage, I represents a phase-modulation machine set output current, represents a voltage-current phase difference, NH represents a sampling point number, represents an instantaneous reactive power value of the ih sampling point; The short-circuit ratio improvement amount is ; ; ; V0 represents a voltage of the PCC bus before a three-phase circuit occurs, I0 represents an effective value of a three-phase short-circuit current when the phase-modulation machine set does not perform reactive power compensation, PDC represents a direct-current power input to the PCC bus, SCR0 represents an initial short-circuit ratio, SCR represents a short-circuit ratio after the phase-modulation machine set is intervened; The filter distortion rate is ; ; represents the effective value of the fundamental current, represents the effective value of the current component with a frequency of h times the fundamental frequency, h represents the harmonic order, and o represents the index of three-phase electricity; represents the maximum value selected from all .

[0039] The dynamic information acquisition module is configured to acquire dynamic parameters on the PCC bus to obtain dynamic performance indexes. The dynamic performance indexes include voltage recovery time, reactive power response time, short-circuit current peak value, and DC power recovery time. The voltage recovery time t v is the time required for the voltage to recover from 0.9 times the rated value to 0.99 times the rated value after the fault is cleared. The reactive power response time t q is the time from the issuance of a step command to the actual reactive power reaching 90% of the command value. The short-circuit current peak value is the maximum instantaneous current contributed by the device during three-phase short-circuit. The DC power recovery time t dc is the time required for the DC power to recover to 99% of the initial value after the AC fault is cleared.

[0040] In the case where the dynamic performance indexes and the static performance indexes are expressed in specific parameters, how to acquire the dynamic performance indexes and the static performance indexes will not be described here.

[0041] The working information acquisition module is configured to acquire the rotor speed, voltage, current, torque, and magnetic flux of the phase-modulating machine group to obtain working performance indexes. The working performance indexes are abnormal features analyzed from original operating parameters. The original operating parameters include rotor speed, voltage, current, torque, and magnetic flux. The voltage and current are measured at the input end of the stator winding. The torque is measured at the motor shaft.

[0042] The above is the specific format and acquisition method of the static performance indexes, dynamic performance indexes, and working performance indexes. The following is the specific structure of the test large model: Referring to Figure 2 and Figure 3 , the test large model includes a performance processing sub-model, a dynamic performance processing sub-model, a static performance processing sub-model, and a performance report generation model.

[0043] The performance processing sub-model generates phase modulator group abnormal state probability information according to the input abnormal features; the performance processing sub-model is a 1-dimensional CNN model + LSTM model, the abnormal features extract effective local features in the CNN model, and then the LSTM captures the correlation between the local features to generate an abnormal state probability vector (abnormal state probability information).

[0044] The dynamic performance processing sub-model generates first hidden information based on the dual-channel input abnormal state probability information and dynamic performance indicators.

[0045] The dynamic performance processing sub-model is a dual-channel fusion network, channel 1 is used to input the abnormal state probability vector, channel 2 is used to input the dynamic information group (dynamic performance indicators), and then the built-in fusion layer (gating fusion mechanism) processes the information input by channel 1 and channel 2 into first hidden information.

[0046] The static performance processing sub-model generates second hidden information based on static performance indicators.

[0047] The static performance processing sub-model is a model structure of multilayer perceptron + encoder, the input static performance indicators learn the nonlinear relationship between the indicators based on MLP, and then the encoder reduces the information to generate second hidden information.

[0048] The performance report generation model generates a performance report of the to-be-tested phase modulator group according to the first hidden information and the second hidden information; the performance report includes the probability of the running stability of the to-be-tested phase modulator group.

[0049] The performance report generation model is a graph neural network model, which constructs a feature relationship graph by the first hidden information and the second hidden information, and then transmits the feature dependence between the GNN messages to output the probability of the running stability of the to-be-tested phase modulator group.

[0050] The performance processing sub-model, the dynamic performance processing sub-model, and the performance report generation model are three independent neural network models, each having independent neural network parameters. During training, they are trained based on their own loss functions and total loss functions.

[0051] Embodiment 2 provides specific data formats based on embodiment 1, and the data formats provided in embodiment 2 are more efficient in training the test large model, and the test large model can more easily find the relationship between the three types of data during the training process.

[0052] Specifically, the static performance indicators obtain the static information group under the single burst signal mode; the dynamic performance indicators obtain the dynamic information group under the multi-burst signal mode; the single burst signal is an amplitude signal or a short circuit signal, and the multi-burst signal is a composite signal of amplitude signal and short circuit signal.

[0053] The static performance index is essentially the limit working capacity of the phase modulator set, so in order to reduce the interference of irrelevant information, only a burst signal is used to test the reactive power regulation range and the reactive power regulation accuracy of the phase modulator set, and only the amplitude signal generated by the step amplitude needs to be provided, that is, the reactive power regulation performance of the phase modulator set is tested. When testing the short-circuit ratio improvement amount and the filter distortion rate of the phase modulator set, only the short-circuit signal provided by the fault injection device is related, so only the short-circuit signal needs to be provided.

[0054] For dynamic performance indicators, it is actually to test the response capacity of the phase modulator set under different working conditions, so it is necessary to comprehensively evaluate the dynamic performance of the phase modulator set. For this purpose, the dynamic performance indicators are tested in a multi-burst signal mode to obtain a dynamic information set.

[0055] Further, a plurality of groups of multi-burst signals are preconfigured, each group of multi-burst signals is loaded in sequence, and the dynamic performance indicators and the working performance indicators are measured synchronously; the collected dynamic performance indicators are taken as a dynamic information set, and the collected working performance indicators are taken as a working performance set.

[0056] When testing the dynamic performance indicators, the dynamic performance obtained under different environments will inevitably differ. Based on this, the application preconfigures a plurality of groups of multi-burst signals, so that when testing, by comparing the change of the dynamic performance indicators under the same environment, the dynamic performance of the phase modulator set can be more accurately compared. At the same time, the dynamic performance indicators and the working performance indicators are measured synchronously. In this way, there is an inherent relationship between the dynamic performance indicators and the working performance indicators, and when the phase modulator set itself has a slight fault, it will inevitably affect the dynamic performance of the phase modulator set. Conversely, if the dynamic performance of the phase modulator set is poor, relevant associated information can also be found from the working performance indicators. In this way, the dynamic information set and the working information set have stronger relevance.

[0057] Further, the plurality of groups of multi-burst signals preconfigured are fault characteristic signals measured in the actual working environment of the phase modulator to be tested.

[0058] In order to ensure that the performance report tested meets the actual application requirements, when configuring the multi-burst signal, the actual environment needs to be considered. Based on this, the plurality of groups of multi-burst signals preconfigured are fault characteristic signals measured in the actual working environment of the phase modulator to be tested.

[0059] The specific way is as follows: a power grid access point similar to the working position of the to-be-tested phase modulation machine group is acquired in advance, the fault signals of the power grid access point are continuously monitored, when a large equipment (for example, a transformer) connected to the power grid fails, or the phase modulation machine group needs to output high-power reactive power or reactive power input, the fault signals of this part are intercepted as fault characteristic signals. The collected fault characteristic signals are grouped to obtain a plurality of groups of multi-burst signals.

[0060] Embodiment 3: The working performance index is actually used to describe the fault characteristics of the phase modulation machine group, but before the phase modulation machine group is put into test, a large amount of quality detection work has been carried out. The phase modulation machine group does not have obvious faults, but only some weak faults, which are generally caused by installation errors of the stator and the rotor, and slight misalignment of the bearings inside the rotor. In actual use, these errors and mistakes are the design redundancies allowed by the assembly work. These weak fault signals do not affect the actual use of the phase modulation machine group. However, when testing the actual performance of the phase modulation machine group, if an accurate performance report needs to be provided, these fault information needs to be considered in the actual operation of the phase modulation machine group. The actual influence of the internal defects of the phase modulation machine group on the performance is analyzed.

[0061] In this way, when the phase modulation machine group responds to the multi-burst signal, the original running parameters are collected, including the rotor speed, voltage, current, torque, and magnetic flux. The abnormal characteristics are extracted from the original running parameters.

[0062] There is little effective information in the original running parameters indicating that there are defects inside the phase modulation machine group, and most of them are normal running signals, which are noise signals for fault detection. Therefore, the application extracts abnormal characteristics by using the following scheme: Based on the particle swarm algorithm, the optimal penalty factor and modal decomposition number are selected from a global perspective, and the original running parameters are decomposed and the abnormal characteristics are reconstructed by using the optimal component factor and modal decomposition number.

[0063] In this scheme, the best penalty factor and modal decomposition number are selected, and then the original running parameters extracted are respectively reconstructed, so that the noise signals can be accurately screened out.

[0064] Further, the extraction method of the abnormal characteristics includes the following steps: S1: obtaining all original running parameters , t = 1, 2, … N; i = 1, 2…5, t is the index of the sampling point, N is the total number of sampling points, and i represents the index of the original running parameter.

[0065] 、 、… These are rotor speed, voltage, current, torque, and magnetic flux, respectively.

[0066] S2: Transfer all original operating parameters VMD signal decomposition was performed separately.

[0067] S3: Use particle swarm optimization algorithm to correct each The penalty factor and the number of mode decompositions during VMD signal decomposition are adjusted to ensure that each... The envelope entropy is minimized when performing VMD signal decomposition.

[0068] S3 requires for each We obtain an optimal penalty factor and mode decomposition number for mode decomposition. For this purpose, we use any one of the following... Taking this as an example, we will explain in detail how to obtain the optimal penalty factor and modality decomposition number.

[0069] refer to Figure 4 S3 includes the following steps: S31: Define the candidate parameter set ; ; Indicates the penalty factor. Represents the modal decomposition number. This indicates the range of values ​​for the penalty factor. Indicates the range of values ​​for the mode decomposition number; S32: Will As the horizontal coordinate in two-dimensional space, As the vertical coordinate in a two-dimensional space, it constitutes the two-dimensional value space of the feasible solution; In a two-dimensional value space, by arbitrarily selecting one value (position), one can determine... Signal decomposition is performed, but determining the optimal value for decomposition is difficult. In practice, the optimal value may exist at any position in two-dimensional space. Therefore, to obtain the optimal penalty factor and mode decomposition number, an algorithm capable of global optimization is needed. Here, "global optimization" means that the particles need to diffuse more uniformly throughout the two-dimensional space and traverse a larger range. Based on this, this application designs a particle swarm optimization algorithm as follows.

[0070] S33: Set the initial number of particles G, the number of iterations M, and the fitness function f(x), and initialize the particle swarm position. Then, iterate over the position of the particles continuously. The fitness function f(x) is to minimize the envelope entropy. S34: Determine if the maximum number of iterations has been reached. If the maximum number of iterations has been reached, stop iterating and output the global optimal solution. If the maximum number of iterations has not been reached, the particle position is updated, and the fitness value of all particles is calculated after the update. If the fitness value is better, the particle position is successfully updated; otherwise, the particle remains in its current position.

[0071] The key to this application lies in selecting the globally optimal solution from the two-dimensional value space, which requires consideration from a global perspective (the particle diffusion direction should not be too concentrated in one direction, nor too disordered). Based on this, the particle diffusion mode is as follows: Determine if the current iteration number m is less than a preset threshold m0; when the iteration number m is less than the threshold m0, particle X... g The updated location is The update method is as follows: ; in, Let represent the final position of particle g in the m-th iteration. This represents the perturbation position of particle g in the m-th iteration. , Let represent the positions of the two randomly selected particles in the t-th iteration. This represents the first random number, used to generate random numbers between [0,1]. Update using the method where the number of iterations m is less than the threshold m0, w d This represents the dynamic weighting coefficient.

[0072] The probability can be calculated using either Formula 1 or Formula 2. The probability calculated using Formula 1 is P, and the probability calculated using Formula 2 is 1-P. Formula 1: ; ; Let represent the global reverse optimal solution at the (m-1)th iteration. This represents the global optimal solution at the (m-1)th iteration; Indicates the upper bound of the search. Indicates the lower bound of the search. b1 represents the second random number, which generates a random number between 0 and 1 for each dimension of the particle, and b1 represents the learning coefficient. Formula 2: ; This represents a 2-dimensional mutation vector. This indicates element-wise multiplication; It is a probability function that generates random numbers only in the x-axis dimension of a binary space.

[0073] ; 1 represents a random value for the vertical axis, & represents a random value for the horizontal axis; & is a random number between (0,1).

[0074] When the number of iterations m is not less than the threshold m0, particle X g The updated location is The update method is as follows: Arbitrarily place two particles X g Particle X j The particles are matched, and their x-coordinates are randomly diffused to obtain the new position of particle g. ; ={ (x), (y)}, (y) represents X g On the y-axis before the update, (x) represents The updated x-axis.

[0075] ; Represents the learning factor. Representing a uniformly distributed random number, controlling the mixing ratio of particles. (x) represents the current position of the g-th particle on the x-coordinate. This represents the current position of the j-th particle on the x-coordinate.

[0076] In this scheme, minimizing the envelope entropy is chosen as the fitness function for updating the particle swarm optimization algorithm because internal faults in the synchronous condenser generally occur periodically along with the relative rotation of the internal rotor and stator. Furthermore, the signal characteristics that appear are all related to rotational irregularities. Therefore, minimizing the envelope entropy can accurately identify periodic abrupt changes.

[0077] The update logic of the particle swarm optimization algorithm in this scheme is as follows: In the early iteration stage, the efficiency of perturbation during particle diffusion is maximized, thereby increasing the randomness of particle diffusion. Simultaneously, to improve the model's convergence efficiency, in Method 2, the ordinate is randomly set to 1 in Equation 2, meaning the ordinate is not perturbed. Thus, in the early stages of particle updates, the perturbation efficiency of the x-coordinate is much higher than that of the ordinate. This is because the range of x-coordinate values ​​is much wider than that of the ordinate, making it relatively easier to find the optimal solution for the ordinate. Therefore, the perturbation range of the ordinate can be reduced to increase the randomness of global diffusion.

[0078] In the later iteration stage, this scheme freezes the iteration of the particle's vertical coordinate and instead randomly pairs two particles to quickly converge the horizontal coordinate, thereby increasing the model's convergence efficiency.

[0079] This application aims to find the optimal VMD signal decomposition scheme, where the VMD signal decomposition scheme is... S4: Adopt When performing VMD signal decomposition, we can obtain a penalty factor that minimizes the envelope entropy and a number of mode decomposition pairs. Perform VMD signal decomposition separately to obtain each IMF signal components; for Use the best penalty factor and mode decomposition number get: ; in, express The k-th IMF component obtained after VMD signal decomposition, where k represents the index of the IMF component, K i express The modal decomposition number.

[0080] S5: Perform feature filtering on the IMF signal components, and then reconstruct the selected IMF components to obtain each... Corresponding reconstruction signal .

[0081] After obtaining the IMF signal components, conventional VMD signal reconstruction methods typically retain the first three or five signal components after mode decomposition, discarding the rest, and then using the remaining components for signal reconstruction. This filtering method may filter out some effective signal components while retaining some noise signals. Therefore, the feature filtering method provided in this application is as follows: S51: Get All The IMF components are used to obtain the IMF component set H; H = {h i ,i=1,2…5}; ; S52: Filter out the first IMF component of each element in the IMF component set H. Establish the comparison component set V; S53: Extract all elements from the IMF component set H except those in component set V, and calculate the correlation between each element in component set V and the component H using the Pearson algorithm. If the average correlation is higher than the preset value, retain the IMF component; otherwise, delete the IMF component. For example, corresponding to h iThere are a total of 10 IMF components. The first IMF component is extracted and added to the comparison component set, while the remaining 9 IMF components need to be filtered. For example, h... i The Pearson coefficients calculated from the second IMF component and the comparison component set V are 0.1, 0.2, 0.1, 0.3, and 0.2, respectively. The average value of these Pearson coefficients is less than the preset value of 0.4, therefore h... i The second IMF component in the array needs to be deleted. And so on, for each h... i The IMF components in the data can all be filtered sequentially using the above method.

[0082] S54: Use the retained IMF components and the IMF components in the comparison component set V to perform VMD algorithm reconstruction to obtain the reconstructed signal. ; express The signal is reconstructed after being processed by the VMD structure. This is an abnormal characteristic.

[0083] For example, h i There are a total of 10 IMF components, of which the 2nd and 5th IMF components have been deleted, so h i Corresponding signal When reconstructing the signal, only the first IMF component, the third and fourth IMF components, and the sixth to tenth IMF components are selected for signal reconstruction to obtain the desired signal. When reconstructing the signal, we strictly adhere to the principle of using the IMF components that we decomposed ourselves.

[0084] The main purpose of this application in using this signal screening method is to find signals that are highly correlated with the first IMF component. Relatively speaking, the higher the correlation between a signal and the signals in the comparison component set V, the stronger the correlation between the signal and the signals in the comparison component set V, and the more likely it is to characterize an anomalous feature signal.

[0085] The above are merely preferred embodiments of this application and are not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A system for coordinated testing of reactive power rejection and step load during synchronous condenser group grid connection, comprising: A power grid simulation device is used to simulate the base current supply of an AC power grid and is connected to the PCC bus via the transmission line impedance. The fault injection device, connected to the PCC bus, is used to simulate a three-phase short circuit generated by equipment connected to the PCC bus in order to generate a short circuit signal. A power step device used to adjust the step amplitude to generate an amplitude signal; Its characteristic is that it further includes: The control module controls the fault injection device and the power step device to randomly inject sudden change signals into the PCC bus; The camera module under test is connected to the PCC bus and responds to the sudden signal injected into the PCC bus by the fault injection device and the power step device to maintain stable power transmission of the PCC bus. The information extraction device extracts the static performance indicators, dynamic performance indicators, and working performance indicators of the phase-shifting camera under test generated on the PCC bus in response to the sudden change signal. The monitoring device inputs static performance indicators, dynamic performance indicators, and operational performance indicators into the large-scale test model to generate a performance report for the camera module under test.

2. The reactive power shedding and step load coordinated testing system for synchronous condenser group grid connection test according to claim 1, characterized in that: Static performance indicators include reactive power regulation range, reactive power regulation accuracy, short-circuit ratio improvement, and filter distortion rate; Dynamic performance indicators include voltage recovery time, reactive power response time, peak short-circuit current, and DC power recovery time; The performance indicators are the abnormal characteristics analyzed from the original operating parameters; The original operating parameters include rotor speed, voltage, current, torque, and magnetic flux.

3. The reactive power shedding and step load coordinated testing system for synchronous condenser group grid connection test according to claim 1, characterized in that: Static performance indicators are obtained from static information groups under single burst signal mode; Dynamic performance indicators acquire dynamic information groups under multiple burst signal modes; A single burst signal is either an amplitude signal or a short-circuit signal, while a multi-burst signal is a composite signal of amplitude and short-circuit signals.

4. The reactive power shedding and step load coordinated testing system for synchronous condenser group grid connection test according to claim 3, characterized in that: Several sets of multi-burst signals are pre-configured, and each set of multi-burst signals is loaded in sequence to simultaneously measure dynamic indicators and working performance indicators. The collected dynamic performance indicators are grouped as dynamic information, and the collected operational performance indicators are grouped as operational performance.

5. The reactive power shedding and step load coordinated testing system for synchronous condenser group grid connection test according to claim 4, characterized in that: The pre-configured sets of multi-burst signals are fault characteristic signals obtained from the actual working environment of the phase converter under test.

6. The reactive power shedding and step load coordinated test system for synchronous condenser group grid connection test according to claim 2, characterized in that: The information extraction device includes: The static information acquisition module is used to collect static parameters on the PCC bus to obtain static performance indicators. The dynamic information acquisition module is used to collect dynamic parameters on the PCC bus to obtain dynamic performance indicators. The working information acquisition module collects the rotor speed, voltage, current, torque, and magnetic flux of the synchronous condenser to obtain working performance indicators.

7. The reactive power shedding and step load coordinated test system for synchronous condenser group grid connection test according to claim 2, characterized in that: The reactive power adjustment range is ( ), This is the minimum value of the reactive power adjustment range. This represents the maximum value within the reactive power adjustment range. reactive power regulation accuracy ; ； ； ； in, This is the reactive power command value. This represents the average value of the actual reactive power. For maximum continuous reactive power output capability, This indicates the PCC bus voltage. This indicates the output current of the synchronous condenser group. This represents the phase difference between voltage and current, and NH represents the number of sampling points. This represents the instantaneous reactive power value at the ih-th sampling point; Short-circuit ratio improvement ; ； ; This indicates the voltage of the PCC bus before the three-phase circuit occurs. This represents the effective value of the three-phase short-circuit current when the synchronous condenser unit is not performing reactive power compensation. This indicates the DC power input to the PCC bus. Indicates the initial short-circuit ratio. Indicates the short-circuit ratio after the synchronous condenser group intervenes; Filter distortion rate ; ； Indicates the effective value of the fundamental current. This represents the effective value of the current component with a frequency h times the fundamental frequency, where h represents the harmonic order and i represents the index of the three-phase electricity. Indicates from all Select the maximum value from the list; Voltage recovery time t v This is the time required for the voltage to recover from 0.9 times the rated value to 0.99 times the rated value after the fault is cleared; Reactive response time t q The time from the issuance of the step command to the actual reactive power reaching 90% of the command value; Peak short-circuit current The maximum instantaneous current contributed by the device during a three-phase short circuit; DC power recovery time t dc The time it takes for the DC power to recover to 99% of its initial value after the AC fault is cleared.

8. The reactive power shedding and step load coordinated testing system for synchronous condenser group grid connection test according to claim 1, characterized in that: The large-scale experimental model includes: The performance processing sub-model generates abnormal state probability information of the camera shifter group based on the abnormal features of the input. The dynamic performance processing sub-model generates the first implicit information based on the dual-channel input abnormal state probability information and dynamic performance indicators; The static performance processing sub-model generates second implicit information based on static performance metrics; The performance report generation model generates a performance report for the camera module under test based on the first implicit information and the second implicit information; the performance report includes the probability of the operational stability of the camera module under test.

9. The reactive power shedding and step load coordinated test system for synchronous condenser group grid connection test according to any one of claims 1 to 8, characterized in that: The optimal penalty factor and mode decomposition number are selected from a global perspective based on the particle swarm optimization algorithm. The original operating parameters are then decomposed using the optimal component factor and mode decomposition number, and the abnormal features are reconstructed from the signal.

10. The reactive power shedding and step load coordinated testing system for synchronous condenser group grid connection test according to claim 9, The extraction of abnormal features includes the following steps: S1: Obtain all raw runtime parameters t=1,2,…N; i=1,2…5, where t is the index of the sampling point, N is the total number of sampling points, and i represents the index of the original running parameters; , … These are rotor speed, voltage, current, torque, and magnetic flux, respectively. S2: Transfer all original operating parameters Perform VMD signal decomposition separately; S3: Use particle swarm optimization algorithm to correct each The penalty factor and the number of mode decompositions during VMD signal decomposition are adjusted to ensure that each... The envelope entropy is minimized when performing VMD signal decomposition. S4: Using various parameters When performing VMD signal decomposition, we can obtain the penalty factor that minimizes the envelope entropy and the number of mode decomposition pairs. Perform VMD signal decomposition separately to obtain each IMF signal components; S5: Perform feature filtering on the IMF signal components, and then reconstruct the selected IMF components to obtain each... Corresponding reconstruction signal , This is an abnormal characteristic.

Images